doi: 10.1111/j.1420-9101.2012.02458.x
Ecological radiation with limited morphological diversification
in salamanders
T. BLANKERS*, D. C. ADAMS & J. J. WIENSà
*Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam, The Netherlands
Department of Ecology, Evolution, and Organismal Biology, Iowa State University, Ames, IA, USA
àDepartment of Ecology and Evolution, Stony Brook University, Stony Brook, NY, USA
Keywords:
Abstract
ecomorphology;
ecological radiation;
morphology;
microhabitat;
phylogenetic comparative method;
Plethodontidae.
A major goal of evolutionary biology is to explain morphological diversity
among species. Many studies suggest that much morphological variation is
explained by adaptation to different microhabitats. Here, we test whether
morphology and microhabitat use are related in plethodontid salamanders,
which contain the majority of salamander species, and have radiated into a
striking diversity of microhabitats. We obtained microhabitat data for 189
species that also had both morphometric and phylogenetic data. We then
tested for associations between morphology and microhabitat categories
using phylogenetic comparative methods. Associations between morphology
and ecology in plethodontids are largely confined to a single clade within
one subfamily (Bolitoglossinae), whereas variation in morphology across
other plethodontids is unrelated to microhabitat categories. These results
demonstrate that ecological radiation and morphological evolution can be
largely decoupled in a major clade. The results also offer a striking contrast to
lizards, which typically show close relationships between morphology and
microhabitat.
Introduction
A major goal of evolutionary biology is to explain the
morphological diversity of plants and animals. In animals, many studies suggest that adaptation to different
microhabitats is a major factor explaining morphological
differences among species within a clade (e.g. Wainwright & Reilly, 1994), especially in the case of adaptive
radiations (e.g. Grant, 1999; Schluter, 2000; Clabaut
et al., 2007; Gavrilets & Losos, 2009). However, a few
studies have also found morphological variation to be
uncorrelated with microhabitat usage, but typically in
smaller clades occupying a limited range of terrestrial
microhabitats (e.g. Vanhooydonck & Van Damme, 1999;
Schulte et al., 2004).
Here, we test whether morphology and microhabitat
usage are related in plethodontid salamanders. PlethoCorrespondence: Thomas Blankers, Institute for Biodiversity and Ecosystem
Dynamics, University of Amsterdam, Sciencepark 904, 1098 XH
Amsterdam, The Netherlands.
Tel.: +1 6319657448 ⁄ +31 641237773;
e-mail: thomasblankers@gmail.com
634
dontidae contains the majority of salamander species
(416 of 608 species; AmphibiaWeb, 2011). Plethodontids
have undergone an impressive ecological radiation into
different microhabitats, and utilize nearly the full range
of microhabitats used by nonmarine and nonvolant
vertebrates, including species that are fossorial, arboreal,
terrestrial, saxicolous, and fully aquatic (e.g. Wake, 1987;
Petranka, 1998).
Many studies have examined the evolutionary morphology and diversification of plethodontids (e.g. Lombard & Wake, 1986; Jockusch, 1997; Adams & Rohlf,
2000; Adams et al., 2009; Kozak et al., 2009). However,
few studies have addressed the relationship between
microhabitat and morphology, and these studies have
generally been confined to single genera (e.g. Desmognathus;
Kozak et al., 2005; Bolitoglossa; Jaekel & Wake, 2007).
Although data on microhabitat use and morphology are
present in the literature for many plethodontid species,
little is known about family-wide relationships between
morphology and microhabitat (Wake, 2009).
Plethodontids are also an intriguing group for studies
addressing ecomorphological relationships because they
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Salamander ecomorphology
(and salamanders in general) are similar in body form to
lizards but only distantly related to them. Like plethodontids, lizards have radiated into a remarkable diversity
of microhabitats (Vitt & Caldwel, 2009). The relationship
between morphology and microhabitat usage in lizards is
relatively well studied, and much of the variation in
lizard morphology (especially the size and shape of the
head, limbs, and tail) is attributable to microhabitat use
in many groups (e.g. Losos, 1990a; Irschick et al., 1997;
Vitt et al., 1997; Wiens et al., 2006; but see Vanhooydonck & Van Damme, 1999; Schulte et al., 2004).
Despite the similarities in body form, there are also major
differences between lizards and plethodontid salamanders in physiology, energetic costs of locomotion, and
behaviour (e.g. plethodontids prefer low temperatures,
have low metabolism, and are lungless; Vitt & Caldwel,
2009).
In this study, we test the relationships between
morphology and microhabitat usage in plethodontid
salamanders. We obtain microhabitat data from the
literature for 189 species for which both morphological
and phylogenetic data are available (Adams et al., 2009).
We then test whether microhabitat is related to morphological variation using phylogenetic comparative
methods. Remarkably, we find that the impressive
ecological radiation of plethodontids is largely decoupled
from variation in morphology (with the exception of a
single clade, the tropical bolitoglossines). These results
suggest that plethodontid salamanders, unlike lizards,
generally retain similar morphologies across a wide range
of microhabitats.
Materials and methods
Phylogeny
The time-calibrated phylogeny used in this study was
taken from Adams et al. (2009). This phylogeny was
constructed by combining a phylogeny for 109 plethodontid species based on nuclear and mitochondrial DNA
data (including all genera and most species in North
America, from Kozak et al., 2009), and a phylogeny for
137 species of tropical bolitoglossines (including all
genera and about a third of the species) from Wiens
et al. (2007), based on mitochondrial DNA data. This
combined phylogeny was then reduced to include only
those species for which morphological data were available (see below). Most analyses are based on the
phylogeny with a root age of 61 Myr (million years).
However, we also tested the robustness of our results to
trees with both younger and older root ages for Plethodontidae (48 and 69 Myr; Adams et al., 2009; note that
these different root ages can influence the relative
lengths of branches, not just the overall depth of the
tree). We note that although different studies have
inferred various root ages for Plethodontidae, the ages of
many major clades within Plethodontidae are similar in
635
our study and a recent multi-locus analysis using
Bayesian divergence-time methods across salamanders
(Zheng et al., 2011). For example, estimates for Spelerpinae, Desmognathus–Phaeognathus, Aneides, and Plethodon
are within 5 Myr or less of each other.
Morphology
Morphological data for seven variables for 189 species
were obtained from Adams et al. (2009). These data are
derived from 1296 museum specimens, including data
from 108 species from Wiens & Hoverman (2008) and
from 81 additional species from Adams et al. (2009). The
sample size varied between 1 and 24 specimens per
species (only adults included), with a mean of 6.7
individuals per species. The morphological variables are
the species means for: snout–vent length (SVL), tail
length (TL; the posterior end of the cloaca to the tail tip),
head length (HL; tip of the snout to the posterior corner
of the jaw opening), body width (BW; measured just
posterior to the forelimb insertion), snout–eye distance
(SE; tip of the snout to posterior corner of the eye
opening), and forelimb and hindlimb length (FLL and
HLL; the posterior insertion of the limb to the tip of the
longest digit, with limbs outstretched). Sexual-size
dimorphism in plethodontids appears to be minor relative to variation in adult body size within species (e.g.
Petranka, 1998), and so data from both sexes were
combined. Prior to all statistical analyses, the variables
were log-transformed to better meet the requirements of
normality and homogeneity.
To remove the effect of size from these linear
measurements, Burnaby’s method (Burnaby, 1966) was
applied, following Adams & Rohlf (2000). The speciesmorphology matrix containing the log-transformed morphological data was projected on the isometric size vector
and returned to the original coordinate system according
to the equation:
XðI lðll0 Þ1 l0 Þ;
where X is the data matrix, I is the identity matrix, and l
represents the isometric growth vector (Burnaby, 1966;
McCoy et al., 2006). The size measure that accompanies
these shape variables is the geometric mean of the linear
measurements for each species (see Mosimann, 1970).
This approach to size and shape is generally preferable
over regression-based methods, which can generate
patterns of false allometry (Mosimann, 1970; Mosimann
& James, 1979). Furthermore, Burnaby’s method is
preferable in this particular case because plethodontid
salamanders are characterized by a high degree of
ontogenetic stasis (i.e. isometry; Jaekel & Wake, 2007;
Adams & Nistri, 2010).
For comparison, the phylogenetic comparative analyses described below were also performed on the nonsizeadjusted data. By performing a PCA (see below) the effect
of size was captured by PC1, while the remaining PCs
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
636
T. BLANKERS ET AL.
represent different relative proportions of the morphological variables (i.e. shape). All statistical analyses were
performed in R version 2.12.1, and, hereafter, only the
specific packages used will be referred to. The R codes for
all analyses are provided in Appendix S1.
Microhabitat
To test the relationship between body form and microhabitat use, species were assigned to microhabitat categories using data from the literature. Microhabitat data
were obtained from various sources, including Wake
(1987), Petranka (1998), McCranie & Wilson (2002),
AmphibiaWeb (2011), and IUCN (2010). The microhabitat categories we primarily used were: arboreal (A),
aquatic (W), semi-aquatic (SW), bromeliad (B), fossorial
(F), saxicolous (S), moss mats (M), and terrestrial (T).
Microhabitat assignment for each species is shown in
Table S1.
The boundaries of each category are explained below,
along with alternative categorizations. Species were
considered arboreal (A) if they typically climb in vegetation higher than 1 m above the ground and are
encountered on leaves, branches, trunks, or epiphytes.
The importance of bromeliad epiphytes (B) for the
ecology of salamanders is sometimes emphasized (e.g.
Wake, 1987). Therefore, we performed some analyses in
which species frequenting bromeliad epiphytes were
assigned to a microhabitat category separate from the
other arboreal species. Moss mats (M) have also been
emphasized as a distinct microhabitat type (e.g. Wake,
1987). Therefore, species that are frequently encountered
on terrestrial moss mats or dense moss on tree trunks
were assigned to the moss mats category in one analysis.
Saxicolous (S) species are those that live in or on rocky
substrates (i.e. within caves or in their twilight zones, on
rock outcrops, or on rocky boulders). Species that
generally live on nonrocky substrates but are sometimes
found under rocks were not categorized as saxicolous.
Species were considered fossorial (F) if they live in
burrows and are rarely encountered outside them.
Species that spend their entire adult life underwater
(e.g. paedomorphic, nonmetamorphosing species) were
assigned to the aquatic (W) microhabitat category. Semiaquatic (SW) microhabitat use was assigned to species
that, during adulthood, routinely spend long periods
under water (e.g. for mating), but also spend considerable time on land. Species that merely live along stream
edges or in splash zones of waterfalls (without submerging in the water for long periods of time) were not
assigned to this category. Finally, terrestrial species
include those that are active primarily on the ground,
including those that spend much of their time under
rocks, logs, debris, or vegetation.
Given that there are potentially many reasonable ways
to assign species to microhabitats, we examined several
alternate coding strategies (scenarios; see descriptions in
Table 1). These scenarios also account for within-species
variation in microhabitat use (Table S1). Species that
occur both in arboreal and bromeliad microhabitats are
assigned to the former category in scenario 1 and to the
latter in scenario 5. Semi-aquatic species are assigned to
the terrestrial category in scenario 2 and to the aquatic
category in scenario 3. Also, fossorial species are treated
separately in the third scenario. In scenario 4 all climbing
species (arboreal, bromeliad, and saxicolous categories)
are combined into one category. Species that were
recorded in terrestrial or arboreal moss mats are assigned
Table 1 Microhabitats included in the nine categorical scenarios.
Scenario
Microhabitats
Description
1
T, W, SW, A (incl. B), S (incl. F)
2
3
4
T (incl. SW), W, A (incl. B), S (incl. F)
T, F, W (incl. SW), A (incl. B), S
T (incl. F), W, SW, A+S+B
5
6
7
T, W, SW, A, B, S (incl. F)
T, W, SW, A (incl. B), S (incl. F), M
T, W, SW, A, B, S (incl. F)
8
T, W, SW, A (incl. B), S (incl. F)
9
T (incl. S, F, SW), W, A (incl. B)
Tests effect of combining bromeliad with arboreal ecotypes, given that living in bromeliads requires
climbing on vegetation
Same as scenario 1, but tests including semi-aquatic species as terrestrial
Same as scenario 2, but tests the effects of combining the semi-aquatic and aquatic categories
Tests importance of climbing, regardless of the substrate, by combining arboreal, bromeliad, and
saxicolous ecotypes.
Tests effect of the bromeliad ecotype on morphology separate from other ecotypes
Tests the consequence of frequenting moss mats*
Reassigns some species of the genus Eurycea to the saxicolous instead of the semi-aquatic
category
Least possible assignments to terrestrial microhabitat: only terrestrial microhabitat specialists were
categorized as terrestrial
Maximal possible assignments to terrestrial microhabitat: all species using terrestrial microhabitats
were categorized as terrestrial
Microhabitat abbreviations are: A (arboreal), B (bromeliad), F (fossorial), S (saxicolous), SW (semi-aquatic), T (terrestrial), W (aquatic), and M
(moss mats).
* All individuals that were assigned to the moss mat ecotype in scenario 6 are assigned to other microhabitat categories in the remaining five
scenarios.
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Salamander ecomorphology
to a moss mat category in scenario 6, and are treated as
either terrestrial or arboreal under other scenarios. The
genus Eurycea contains many species that occur in
saxicolous microhabitats in addition to aquatic microhabitats and these species are classified as saxicolous in
scenario 7. In scenario 8, species that occur in both
arboreal and terrestrial microhabitats are treated as
arboreal and species that occur in both saxicolous and
terrestrial microhabitats are categorized as saxicolous. In
contrast, in scenario 9, these two sets of species are both
categorized as terrestrial.
Taken together, these various coding strategies generally account for species that occur in two or more
microhabitats (i.e. under one coding strategy, they are
coded as occurring in one of their microhabitats, and are
coded as occurring in another under an alternative
strategy). Nevertheless, a ‘default’ microhabitat was still
needed in some cases. This default microhabitat category
was chosen as the microhabitat in which the species was
most frequently encountered, based on the references
cited (the predominant microhabitat is listed in the first
column in Table S1). However, in a minority of cases, it
was not clear which microhabitat was predominant, and
these were coded somewhat arbitrarily. But again, even
these cases are accounted for by our various coding
strategies. Among the 189 species, there are three species
that occur in multiple microhabitats (and in which some
of the microhabitats used by the species are not included
in the coding schemes), but in all three of these species
one of the microhabitats is more commonly used than
the others, making the coding relatively unambiguous.
We also performed a series of analyses in which we
tested specific microhabitats against all other microhabitats (treating microhabitat as binary). These included:
arboreal and bromeliad (scenario 10), terrestrial and
semi-aquatic (scenario 11), aquatic and semi-aquatic
(scenario 12), saxicolous and fossorial (scenario 13),
bromeliads (scenario 14), and a combination of arboreal
and saxicolous (scenario 15). For species occurring in
multiple microhabitats, any occurrence in the microhabitat in question was coded as present.
Ecological radiation
In order to visualize the microhabitat variation among
species on the tree, we used maximum likelihood
reconstruction (e.g. Schluter et al., 1997; Pagel, 1999).
This was implemented in Mesquite version 2.74 (Maddison & Maddison, 2010), with the Markov k-state, onerate model (a single rate for all transitions between all
states, the most practical method given the large number
of states). To better visualize the variation among species,
we included all eight microhabitat categories as character
states, except for the moss mats category. We consider
moss mats as a secondary microhabitat, utilized by
species in addition to either terrestrial or arboreal
microhabitats.
637
Principal components analysis
After size-correction, species means for the morphological variables were included in a principal components
analysis (PCA), using the pcaMethods package (Stacklies
et al., 2007). The PCA was performed on the covariance
matrix and six axes were retained. Species scores along
each PC axis were obtained and used in the phylogenetic
comparative analyses. Similarly, a PCA was performed on
the covariance matrix of the seven log-transformed (but
not size-adjusted) morphological variables for comparison with the results of the analyses on the size-corrected
variables.
Phylogenetic comparative analyses
To address the relationship between morphology and
microhabitat, we used phylogenetic generalized least
squares (PGLS; Grafen, 1989; Martins & Hansen, 1997).
In PGLS, the original (nonphylogenetically corrected)
data are converted by multiplication with the variance–
covariance matrix derived from the tree (e.g. using the
vcv.phylo function in the ape package; Paradis et al.,
2004). This can be done using different models of
evolution, including Brownian motion (BM) and
Ornstein–Uhlenbeck (OU).
The effects of size were represented by the geometric
mean vector obtained after the Burnaby size-correction,
and PC1–PC6 represented shape (Table 2). When no sizecorrection was applied prior to the PCA, PC1 was
considered to represent size, since the loadings for all
the morphological variables were roughly equal on this
axis (Jolicoeur, 1963). The remaining six PCs show
different patterns of variation in different sets of morphological variables, and are considered to represent
variation in shape.
Using the size-corrected variables, we tested if body
size and shape (size-vector and multivariate shape;
dependent variables) are each related to microhabitat
use (independent variables), using the 15 scenarios for
categorizing microhabitat use (Table 1). For shape, we
first examined the effect of microhabitat use on PC1–PC6
simultaneously by performing a multivariate PGLS for
the microhabitat scenarios (using the nine categorical
scenarios only). Presently, no R packages are available
that perform multivariate PGLS, as the summary.manova
routine does not retain covariance structures passed to it.
Therefore we wrote a code for PGLS that handles a
multivariate Y variable (see R-codes in Appendix S1).
Estimates of lambda (phylogenetic signal) and r2 were
not obtained, as neither has been developed for multivariate models.
In addition, we performed an exploratory analysis
using each of the six PC-axes separately as response
variables in a PGLS with the forward selection scheme for
the microhabitat scenarios. To find the best possible
evolutionary model for the (univariate) PGLS, we first
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
638
T. BLANKERS ET AL.
used PGLS implemented in CAIC (Orme et al., 2009) to
obtain the maximum likelihood estimation of the
strength of phylogenetic signal (k; Pagel, 1999). We also
used nlme (Pinheiro et al., 2011) to find the expected
variance-covariance matrix under a given evolutionary
model (i.e. BM or OU). Using other PGLS models (e.g.
from the caper package; Orme et al., 2011) did not affect
the results and are therefore not included in the analyses.
The likelihood of the data under the BM and OU models
for each PC was estimated using the geiger package
(Harmon et al., 2008). The actual covariance structure
under each model was obtained in the PGLS analysis (see
R-codes in Appendix S1). For each scenario, the most
likely model was selected using the Akaike Information
Criterion (Akaike, 1983). One model was preferred over
another if the AIC scores differed by at least four
(Burnham & Anderson, 2002). When AIC scores differed
by < 4, we performed the analysis with both models.
There was only one instance in which we found two
models with AIC scores that differed by < 4 (see ‘Results’
section). Here, the models showed similar results and we
present only the results of one of the two models in the
‘Results’ section.
For the size-uncorrected variables, we analysed the
relationship between microhabitat use and morphology
using the microhabitat classification schemes in scenarios
1–6. The first six scenarios together test the effects of all
microhabitat categories included in our analysis (and
gave results very similar to those for the size-corrected
data; see ‘Results’ section) and were therefore considered
sufficient for robustness analyses, rather than testing all
possible scenarios. Similarly, we used the most likely
model and the first six scenarios to test the effect of
different root ages (i.e. 48 and 69 Myr) for the timecalibrated tree. The results were compared with the
results obtained from the original analysis with sizecorrected variables and a root age of 61 Myr.
Sequential Bonferroni correction was applied to the
results following Rice (1989). This correction redefines
the significance threshold as P £ a ⁄ (1 + k - i), for the ith
P-value of k tests. Only P-values that remained below the
significance threshold after sequential Bonferroni correction were considered significant.
Given that ecomorphological associations may differ
between clades and regions, temperate and tropical
plethodontids were analysed separately in some analyses.
We assigned all tropical bolitoglossines (106 species;
including all tropical genera) to a tropical subgroup. The
remaining 83 species were assigned to a temperate
subgroup (including one genus of Bolitoglossinae, one
of Hemidactyliinae, four of Spelerpinae, and seven of
Plethodontinae). Note that the tropical bolitoglossines are
a monophyletic group whereas the temperate subgroup is
not, but this should not adversely influence the analysis.
The PCA and PGLS analyses described above were
performed on these subgroups separately. However,
fewer scenarios were used in these PGLS analyses, as
not all microhabitats were represented in both subgroups.
For the temperate subgroup we analysed scenarios 1–4, 6,
10–13, and 15. For the tropical subgroup the analysis was
restricted to scenarios 1, 4, 5, 10, 13, and 15. For both
groups we considered the effects of microhabitat partitioning on both shape (i.e. PC1–PC6 in the case of
Burnaby size-corrected variables and PC2–PC7 in the case
of size-uncorrected variables) and size (i.e. the geometric
mean derived from the Burnaby size-correction and PC1
for the size-uncorrected data).
Results
Ecological radiation
We reconstructed the evolution of microhabitat use
across plethodontids (Fig. 1). Transitions between microhabitats occur frequently. Arboreal and bromeliad species
are most frequent in the tropical clade (species 2–107 in
Fig. 1). In contrast, aquatic, semi-aquatic, and saxicolous
species are more common among temperate species
(species 1, and 108–189 in Fig. 1).
Principal components analysis
Loadings for the morphological variables on the sizecorrected PCs are shown in Table 2. With the exception
of PC7 (which has zero-variance after size-correction)
each PC reflects the influence of different sets of
morphological variables. Larger values on PC1 indicate
increasing tail length and decreasing limb length. Larger
values on PC2 reflect increasing body width and decreasing limb length. PC3 reflects head length relative to other
body proportions. Larger values on PC4 reflect increased
snout–eye distance. PC5 reflects contrasts of snout–vent
length with head length and body width. PC6 reflects the
contrasting lengths of the forelimb and hindlimb. The
loadings for the morphological variables on the PCs were
generally similar when the PCA was performed on the
tropical and temperate subgroups separately (Tables S7
and S10). Plotting PC scores for all species for the axes
that explain the most shape variation shows that species
do not cluster distinctly based on their microhabitat
assignment and instead show extensive overlap with
species utilizing different microhabitats (Fig. 2).
Phylogenetic comparative analyses
For the multivariate analysis, the BM model is preferred
over the OU model, both for the analysis of all Plethodontidae and of the temperate and tropical subgroups
(Table S2). For the univariate analyses, the model with
estimated lambda (the strength of the phylogenetic
signal; implemented in CAIC) is generally favoured, with
the exception of the analysis of the entire family on PC4,
temperate species on PC5, and tropical species on PC3–
PC6. In the former two exceptions, the fixed-OU model is
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Salamander ecomorphology
639
Fig. 1 Evolution of microhabitat use across the phylogeny of plethodontid salamanders. Species were assigned to one of the seven
microhabitat categories following scenario 5 (with fossorial as an
additional separate category). Ancestral states were estimated using
maximum likelihood reconstruction under the one-rate model.
Branch colours represent microhabitat use of extant species and
inferred ancestors. Branches with more than one colour indicate those
where there is not significant support for one state over another
(difference in log likelihood units for alternate states is < 2). Arrows
indicate species shown in the photographs (each representing one of
the 15 clades used in Adams et al., 2009). See acknowledgments for
photo credits. Every tenth species is numbered. 1: Hemidactylium
scutatum, 2: Bolitoglossa schizodactyla, 3: B. colonnea, 4: B. adspersa, 5:
B. medemi, 6: B. altamazonica, 7: B. peruviana, 8: B. palmata, 9: B. biseriata,
10: B. sima, 11: B. gracilis, 12: B. pesrubra, 13: B. subpalmata, 14: B.
cerroensis, 15: B. minutula, 16: B. marmorea, 17: B. carri, 18: B. conanti, 19:
B. diaphora, 20: B. dunni, 21: B. morio, 22: B. celaque, 23: B. synoria, 24:
B. decora, 25: B. longissima, 26: B. porrasorum, 27: B. franklini, 28: B.
lincolni, 29: B. dofleini, 30: B. engelhardti, 31: B. helmrichi, 32: Bolitoglossa
rostrata, 33: B. hermosa, 34: B. riletti, 35: B. macrinii, 36: B. flaviventris, 37:
B. platydactyla, 38: B. mexicana, 39: B. striatula, 40: B. yucatana, 41:
B. lignicolor, 42: B. odonnelli, 43: B. occidentalis, 44: B. rufescens, 45:
B. hartwegi, 46: Pseudoeurycea naucampatepetl, 47: P. gigantea, 48:
P. boneti, 49: P. bellii, 50: P. cephalica, 51: P. galeanae, 52: P. scandens, 53:
Ixalotriton niger, 54: I. parvus, 55: Parvimolge townsendi, 56: Lineatriton
lineola, 57: Pseudoeurycea firscheini, 58: P. leprosa, 59: P. lynchi, 60: P.
nigromaculata, 61: P. mystax, 62: P. werleri, 63: P. juarezi, 64: P. saltator,
65: P. unguidentis, 66: P. altamontana, 67: P. robertsi, 68: P. longicauda, 69:
P. melanomolga, 70: P. gadovii, 71: P. cochranae, 72: P. anitae, 73: P. smithi,
74: P. rex, 75: P. exspectata, 76: P. brunnata, 77: P. goebeli, 78:
Chiropterotriton multidentatus, 79: C. arboreus, 80: C. dimidiatus, 81:
C. lavae, 82: C. priscus, 83: C. chondrostega, 84: C. magnipes, 85: Thorius
dubitus, 86: T. troglodytes, 87: Bradytriton silus, 88: Oedipina pacificensis,
89: O. uniformis, 90: O. cyclocauda, 91: O. pseudouniformis, 92: O. poelzi,
93: O. savagei, 94: O. complex, 95: O. parvipes, 96: O. elongata, 97:
O. gephyra, 98: Nototriton guanacaste, 99: N. abscondens, 100: N. picadoi,
101: N. richardi, 102: N. barbouri, 103: Dendrotriton cuchumatanus, 104:
D. rabbi, 105: Nyctanolis pernix, 106: Cryptotriton alvarezdeltoroi, 107:
C. nasalis, 108: Batrachoseps wrighti, 109: B. attenuatus, 110: B. major,
111: Stereochilus marginatus, 112: Gyrinophilus porphyriticus, 113:
Pseudotriton montanus, 114: P. ruber, 115: Eurycea rathbuni, 116: E. nana,
117: E. sosorum, 118: E. neotenes, 119: E. tridentifera, 120: E. cirrigera E,
121: E. bislineata, 122: E. junaluska, 123: E. wilderae J, 124: E. cirrigera L,
125: E. wilderae M, 126: E. lucifuga, 127: E. longicauda, 128: E.
quadridigitata, 129: E. multiplicata, 130: E. tynerensis, 131: E. spelaeus,
132: Plethodon vandykei, 133: P. idahoensis, 134: P. neomexicanus, 135:
P. elongatus, 136: P. vehiculum, 137: P. dunni, 138: P. cinereus, 139:
P. shenandoah, 140: P. nettingi, 141: P. hubrichti, 142: P. richmondi, 143:
P. hoffmani, 144: P. serratus, 145: P. wehrlei, 146: P. punctatus, 147:
P. welleri, 148: P. angusticlavius, 149: P. ventralis, 150: P. dorsalis, 151:
P. glutinosus, 152: P. mississippi, 153: P. albagula, 154: P. grobmani, 155:
P. montanus, 156: P. amplus, 157: P. meridianus, 158: P. shermani, 159:
P. cheoah, 160: P. chlorobryonis, 161: P. teyahalee, 162: P. cylindraceus,
163: P. caddoensis, 164: P. ouachitae, 165: P. metcalfi, 166: P. jordani, 167:
P. kentucki, 168: P. yonahlossee, 169: Karsenia koreana, 170: Hydromantes
italicus, 171: H. platycephalus, 172: Ensatina eschscholtzii, 173: Aneides
aeneus, 174: A. hardii, 175: A. lugubris, 176: A. ferreus, 177: A.
flavipunctatus, 178: Phaeognathus hubrichti, 179: Desmognathus wrighti,
180: D. aeneus, 181: D. quadramaculatus D, 182: D. marmoratus B, 183:
Desmognathus imitator, 184: D. ochrophaeus, 185: D. welteri, 186: D. fuscus
B, 187: D. monticola B, 188: D. brimleyorum and 189: D. auriculatus.
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
640
T. BLANKERS ET AL.
Table 2 Eigenvalues, percentage of variance explained, and loadings for the seven size-adjusted morphological variables on the six
PCs.
PC1
Eigenvalue
% variance
explained
SVL
TL
HL
SE
BW
FLL
HLL
PC2
0.124
63
0.23
0.81
)0.08
)0.14
)0.12
)0.36
)0.34
0.032
18
0.17
)0.23
)0.16
0.29
0.71
)0.30
)0.46
PC3
PC4
0.017
10
)0.27
0.27
)0.71
)0.21
0.36
0.37
0.19
0.010
5
)0.31
0.13
)0.25
0.83
)0.34
0.04
)0.12
PC5
0.006
3
)0.78
0.23
0.48
)0.06
0.31
)0.12
)0.06
PC6
0.003
1
0.02
0.00
0.16
)0.11
)0.08
0.70
)0.69
Bold values indicate the three variables with the strongest loadings
on that axis.
favoured over the two other models. For the PGLS
models of the tropical species on PC3–PC6 the AIC scores
of the estimated lambda model and the fixed-OU model
are different by < 4. The outcomes of these two models
are, however, qualitatively similar and we only show the
results of the fixed-OU model. In Table S2, the AIC scores
representing the fit of the PGLS models using the first
microhabitat coding strategy (scenario 1) are shown. The
models have similar fit under the 14 other coding
strategies, hence showing these results is redundant.
In the multivariate analysis, we found a statistically
significant relationship between morphology and microhabitat use for all nine categorical (nonbinary) scenarios
for describing microhabitat variation (P-values range
from < 0.0001 to 0.0361; Table 3). In contrast, we found
very few significant associations between microhabitat
use and the individual PCs representing shape (Tables 4
and 5). Only the relationship between microhabitat use
and PC5 (snout–vent length vs. head length and body
width) is significant (P ranges from 0.0001 to 0.0183;
Table 4). This relationship is consistent with the observation that terrestrial, aquatic, and saxicolous species
have a more slender body (relative to their snout–vent
length) as compared to species using other microhabitats
(Table S3). Excluding PC5 from the multivariate analysis
(hence using PC1–PC4 and PC6 as a multivariate
response variable) results in nonsignificant P-values for
the relationship between microhabitat and shape
(Table S4). Regardless of the strategy of microhabitat
assignment, the PGLS models show that microhabitat
explains only a small proportion of the overall morphological variation (r2 < 0.13 for all scenarios).
When microhabitat use is treated as a binary
variable (scenarios 10–15; Table 5) we find that the
‘climber’ (saxicolous + arboreal), bromeliad, and arboreal
microhabitats are associated with PC4 (bromeliad-dwellers and arboreal and saxicolous climbers have a larger
snout–eye distance) and PC5 (saxicolous species have
(a)
(c)
(b)
(d)
Fig. 2 Morphospace distribution from principal components analysis (PCA) on Burnaby size-corrected morphological variables for 189
plethodontid salamander species. The species scores on PC1 and PC2 (a and d), PC1 and PC3 (b), and PC1 and PC4 (c) are displayed. In (a), (b),
and (c), colours are based on microhabitat assignment, but in (d), scores are coloured according to the geographic region of the species.
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Salamander ecomorphology
Table 3 Relationship between microhabitat categories and morphology, including all PCs simultaneously.
AIC
All species (n = 189)
Scenario 1
)5788.6
Scenario 2
)5786.0
Scenario 3
)5786.3
Scenario 4
)5783.9
Scenario 5
)5799.7
Scenario 6
)5801.6
Scenario 7
)5815.8
Scenario 8
)5798.6
Scenario 9
)5781.4
F
numDf
denDf
P
1.6606
1.7784
1.6448
1.5932
1.8352
1.7461
2.2207
1.6579
1.8674
30
24
30
24
36
36
36
30
18
915
732
915
732
1098
1098
1098
915
549
0.0148
0.0100
0.0165
0.0361
0.0021
0.0045
5.63 · 10)5
0.0151
0.0163
Table 4 Relationship between microhabitat categories and morphology using separately analysed PCs, including all species.
Scenario
PC1
PC2
PC3
r2
k
F4,184
P
0.011
0.964
0.528
0.716
0.032
0.868
1.504
0.203
0.015
0.647
0.703
0.591
r2
k
F3,185
P
0.011
0.964
0.692
0.558
0.012
0.873
0.773
0.511
0.010
0.659
0.613
0.607
r2
k
F4,184
P
0.005
0.962
0.254
0.907
0.047
0.846
2.251
0.065
0.012
0.655
0.554
0.697
r2
k
F3,185
P
0.011
0.962
0.700
0.553
0.033
0.865
2.115
0.100
0.015
0.649
0.955
0.415
r2
k
F5,183
P
0.012
0.965
0.452
0.811
0.034
0.870
1.274
0.277
0.050
0.659
1.946
0.089
r2
k
F5,183
P
0.011
0.962
0.399
0.849
0.034
0.866
1.302
0.265
0.021
0.653
0.772
0.571
r2
k
F5,183
P
0.005
0.968
0.184
0.968
0.020
0.877
0.752
0.585
0.060
0.717
2.347
0.043
r2
k
F4,184
P
0.012
0.966
0.575
0.681
0.033
0.871
1.547
0.191
0.025
0.664
1.176
0.323
r2
k
F2,186
P
0.001
0.963
0.121
0.886
0.011
0.863
1.064
0.347
0.010
0.648
0.916
0.402
PC4*
PC5
PC6
Size
2.2589
0.0645
0.106
0.623
5.435
0.0004
0.052
0.674
2.515
0.043
0.005
0.899
0.243
0.914
2.7442
0.0445
0.106
0.624
7.279
0.0001
0.014
0.717
0.895
0.445
0.005
0.905
0.305
0.822
2.3223
0.0584
0.104
0.618
5.356
0.0004
0.023
0.710
1.088
0.364
0.007
0.908
0.340
0.851
2.7338
0.0451
0.040
0.648
0.055
0.055
0.052
0.672
3.370
0.020
0.001
0.906
0.042
0.989
1.7898
0.1170
0.128
0.635
5.350
0.0001
0.067
0.689
2.609
0.026
0.012
0.896
0.437
0.822
1.4696
0.2018
0.071
0.594
2.802
0.018
0.058
0.691
2.237
0.053
0.020
0.909
0.763
0.577
1.9084
0.1108
0.124
0.648
5.187
0.0002
0.039
0.715
1.488
0.196
0.015
0.901
0.549
0.739
3.0786
0.0484
0.084
0.589
4.242
0.003
0.056
0.667
2.719
0.031
0.005
0.905
0.225
0.924
1.7506
0.1252
0.088
0.631
8.973
0.0002
0.011
0.726
1.055
0.350
0.001
0.907
0.123
0.884
1
2
3
Temperate species (n = 83)
Scenario 1
)2655.7
Scenario 2
)2651.2
Scenario 3
)2669.2
Scenario 4
)2653.0
Scenario 6
)2656.7
0.7153
0.4765
0.7994
0.7913
0.7954
30
24
30
24
36
385
308
385
308
462
0.8670
0.9838
0.7674
0.7478
0.7978
Tropical species
Scenario 1
Scenario 4
Scenario 6
3.3153
2.8052
3.4279
18
24
18
300
400
300
9.42 · 10)6
1.76 · 10)5
0.0001
(n = 106)
)2745.6
)2742.3
)2746.1
641
PGLS was performed on PC1–PC6 after size-correction, with
categorical microhabitat use as the independent variable. Different
scenarios indicate different ways of describing microhabitat variation
(see Table 1). Results are shown separately for the entire family, the
temperate species, and the tropical species. AIC scores (AIC),
F-statistics (F), degrees of freedom (numDf for numerator and denDf
for denominator degrees of freedom) and P-values (P) are shown.
Significant results after sequential Bonferroni correction are shown
in boldface type.
relatively higher values for body width and head length,
whereas bromeliad-dwellers and arboreal species have
lower values for those variables). The only associations
that are significant after Bonferroni correction are
between the bromeliad microhabitat category and PC5
(scenario 13: F1,187 = 21.957; P < 0.0001), and microhabitat categories that involve climbing (i.e. arboreal
[incl. bromeliad] and saxicolous) and PC4 (scenario 14:
F1,187 = 8.5602; P = 0.0039; scenario 15: F1,187 = 9.8756;
P = 0.0019). Comparing these results to the raw data
(Table S1), we find that the observed association with PC5
is attributable to differences in body width relative to
snout–vent length, rather than to differences in head
length.
Comparable results are found when analysing the sizeuncorrected variables (Table S5). After sequential Bonferroni correction, we find only PC6 (body width and
head length relative to snout–vent length; corresponding
to PC5 in the analysis of size-corrected data) to be
significantly associated with microhabitat use (P ranges
between 0.0001 and 0.0207, d.f. as in Table 4). As in the
analysis of size-corrected variables, relatively little of the
4
5
6
7
8
9
PGLS was performed on all seven PCs after size-correction,
with categorical microhabitat use as the independent variable.
Different scenarios indicate different ways of describing
microhabitat variation (see Table 1). Percentage of variance
explained (r2), phylogenetic signal (k), F-statistics (F with degrees
of freedom in subscript), and P-values (P) are shown. Significant
results after sequential Bonferroni correction are shown in
boldface type.
* Analysis performed using (fixed-OU) model from R-package ‘nlme’
that returns no values for r2 and k.
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
642
T. BLANKERS ET AL.
Table 5 Relationship between binary microhabitat categories and
morphology among all 189 sampled plethodontid species.
Scenario PC1
10 (terrestrial)
0.004
r2
PC2
0.001
PC3
PC4*
PC5
0.000
0.018
k
0.965 0.881 0.675
0.636
r
0.024 )0.010 0.002 )0.0175 )0.019
0.738 0.241 0.011 1.5156 3.451
F1,187
P
0.392 0.624 0.916 0.2198 0.065
11 (aquatic)
0.008 0.012 0.009
0.007
r2
k
0.962 0.875 0.656
0.611
r
)0.139 0.102 )0.070 )0.0157 )0.036
1.570 2.275 1.729 0.1483 1.321
F1,187
P
0.212 0.133 0.190 0.7006 0.252
12 (saxicolous)
0.002 0.000 0.000
0.022
r2
k
0.965 0.879 0.679
0.579
r
)0.027 )0.008 0.006 0.0074 )0.033
0.321 0.063 0.047 0.1301 4.298
F1,187
P
0.571 0.803 0.829 0.7187 0.040
13 (bromeliad)
r2
0.001 0.001 0.009
0.105
k
0.964 0.880 0.677
0.648
r
)0.018 )0.011 )0.029 0.0399 0.061
F1,187
0.273 0.187 1.658 4.6126 21.957
P
0.602 0.666 0.199 0.0330 5.36 ·
10)6
14 (climber)
r2
0.004 0.001 0.004
0.002
k
0.963 0.879 0.687
0.611
r
)0.028 )0.007 0.016 0.0446 0.007
0.793 0.114 0.687 8.5602 0.411
F1,187
P
0.374 0.736 0.408 0.0039 0.522
15 (arboreal)
0.002 0.000 0.003
0.021
r2
k
r
F1,187
P
0.963
)0.022
0.351
0.555
0.879
0.007
0.070
0.791
0.684
0.016
0.500
0.480
0.0584
9.8756
0.0019
0.605
0.028
4.094
0.044
PC6
Size
0.731
0.002
0.059
0.809
4.411 ·
10)5
0.906
)0.009
0.008
0.928
0.013
0.721
0.030
2.375
0.125
0.000
0.907
0.096
0.073
0.788
0.001
0.725
0.005
0.223
0.637
0.002
0.904
0.099
0.373
0.542
0.000
0.004 0.008
0.727 0.899
0.007 )0.158
0.794 1.581
0.374 0.210
0.000
0.727
0.001
0.045
0.833
0.001
0.906
0.052
0.228
0.634
0.001
1.09 ·
10)5
0.906
)0.006
0.002
0.964
0.733
)0.003
0.164
0.686
PGLS was performed on all seven PCs after size-correction, with
binary categorized microhabitat use as the independent variable.
Different scenarios indicate different ways of describing microhabitat
variation (see Table 1). Percentage of variance explained (r2),
phylogenetic signal (k), regression coefficients (r), F-statistics
(F, degrees of freedom in subscript), and P-values (P) are shown.
Significant results after sequential Bonferroni correction are shown
in boldface type.
* Analysis performed using (fixed-OU) model from R-package ‘nlme’
that returns no values for r2 and k.
overall variation in morphology is explained by microhabitat (r2 < 0.12 for all scenarios).
Using phylogenetic trees with younger and older root
ages (i.e. 48 and 69 Myr) did not influence the overall
results of our analyses. Similar to our findings described
above, we find only PC5 to be significantly associated
with microhabitat use (Tables S6 and S7).
Tropical and temperate subgroups
For the temperate species, very few significant relationships between microhabitat use and body shape are
evident and no statistically significant association
between microhabitat use and size was found (Tables 3,
S9 and S10). The multivariate analysis revealed no
significant associations between microhabitat and morphology in temperate species (Table 3). Similarly, in all
but one scenario, no significant associations are found
between microhabitat use and the individual shape PCs.
The single exception involves the association between
microhabitat and PC2 for scenario 3 (F4,78 = 6.81;
P < 0.0001; Table S9). However, for this scenario, in
which fossorial microhabitat use is considered separately
(instead of combined with, for example, terrestrial
microhabitat use), only one species was categorized as
fossorial in the temperate subgroup (Table S1).
For the tropical subgroup, results are very similar to
those across the family. The multivariate analysis shows a
strong relationship between microhabitat use and shape
for all scenarios (P-values from < 0.0001 to 0.0010;
Table 3), but not for the microhabitat-size association.
Exploring the PCs individually reveals a significant
association between PC3 (head length and snout–eye
distance vs. body width) and microhabitat use (categorical scenarios 1 and 4 and binary scenario ‘bromeliad’
[13]; P-values range between 0.0014 and 0.0233). Also,
microhabitat use (all categorical scenarios and binary
scenarios ‘terrestrial’ [10] and ‘bromeliad’ [13]) is significantly correlated with PC5 (body width relative to
snout–vent length; P-values range between 0.0061 and
0.0176; Table S12). Therefore, the relationship between
morphology and microhabitat seems to be related to
differences in snout–vent length and body width between arboreal (broader) and terrestrial (slender) species.
In summary, our results show that there is little
relationship between morphology and microhabitat in
plethodontid salamanders, and the relationship that is
present is driven primarily by shape variation in the
tropical clade alone.
Discussion
A major paradigm in evolutionary biology and ecology is
that the morphology of species often reflects adaptation
to the microhabitats in which they occur (e.g. Wainwright & Reilly, 1994). This relationship has been shown
in numerous studies, particularly those focusing on
adaptive radiation (e.g. Schluter, 2000). Our results from
plethodontid salamanders offer an interesting counterexample. Although we do find a significant relationship
between variation in morphology and microhabitat
when including all 189 species, this relationship seems
to be largely attributable to variation in a few morphological traits and only within the relatively recent but
species-rich tropical clade (supergenus Bolitoglossa). In
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Salamander ecomorphology
contrast, among the temperate species (which include all
four subfamilies and nearly all microhabitats), the relationship between morphology and microhabitat is largely
absent, both when shape is included as a multivariate
dependent variable and as separate univariate variables.
Additionally, across the entire family, most individual
PCs representing different aspects of shape show no
relationship with microhabitat use (and those that do
explain very little of the overall morphological variation,
with r2 < 0.13). Also, size and microhabitat are unrelated
across all plethodontids and in separate analyses of the
temperate and tropical subgroups. In short, variation in
the tropical clade seems to explain most of the observed
relationships between morphology and microhabitat
categories in plethodontids, despite a remarkable ecological radiation among the temperate species (Fig. 1),
encompassing nearly all possible microhabitats for nonmarine vertebrates (e.g. aquatic, arboreal, fossorial,
saxicolous, terrestrial).
In theory, the limited relationships observed between
morphology and microhabitat could be an artifact of any
one of several factors, but most of these seem unlikely.
One of these factors is measurement error. However, the
morphometric data of Adams et al. (2009) show strong
repeatability between measurers (r = 0.99), suggesting
that the data are generally accurate. Furthermore, the
morphological shape variables generally show a strong
phylogenetic signal (typically k 0.9 in the univariate
analyses [Table 5]). This result indicates that the morphological variation tracks the phylogeny (and is therefore nonrandomly distributed among species), but
nevertheless fails to track changes in microhabitat.
Another possibility is that our study had insufficient
statistical power. However, we utilized a large sample size
of species (n = 189 for the entire family and n = 83 and
106 for the temperate and tropical subgroups, respectively). Additionally, even though only 45% of plethodontid species were included, our sampling includes
nearly all genera (with the exception of the recently
described Urspelerpes; AmphibiaWeb, 2011). Thus, it
seems likely that most of the major variation in microhabitat and morphology within plethodontids was
included. We acknowledge that there could (in theory)
be stronger ecomorphological relationships in salamanders outside plethodontids. However, except for a few
elongate, paedomorphic genera in other families (e.g.
Amphiuma, Proteus, Pseudobranchus, Siren), salamander
body forms in other families are generally similar to those
in plethodontids (e.g. Wiens & Hoverman, 2008). There
are also many potential ambiguities in assigning species
to microhabitats, but we employed a large number of
alternative coding schemes, none of which show strong
relationships between morphology and microhabitat.
However, we acknowledge that there could be other
aspects of microhabitat that are important but not
reflected in our coding. Nevertheless, it seems that
morphology should generally track these coarse micro-
643
habitat categories rather than finer distinctions within
them (e.g. we postulate that the important distinction for
morphology should be between microhabitats that
require salamanders to climb vs. burrow vs. swim, and
that minor differences in, for example, arboreal microhabitats should be less relevant for explaining morphological variation). Also, although it would be interesting
to analyse the effect of within-category variation on
morphology, this may require additional data on microhabitat utilization beyond what are currently available
for most species.
Perhaps the most important limitation of our study is
that it can only reflect the morphological variables that
we measured. Nevertheless, studies of lizards (which are
similar in general body form) suggest that microhabitat
can strongly influence the relative proportions of the
head, limbs, trunk, and tail (e.g. Losos, 1990a,b; Vitt
et al., 1997; Irschick & Losos, 1998; Glor et al., 2003;
Wiens et al., 2006). Thus, the variables that we used are
expected to be particularly important for locomotion in
different microhabitats. Nevertheless, there may still be
important variation in other morphological variables we
did not take into account in the present study that may
be related to microhabitat (e.g. foot shape and interdigital
webbing; Jaekel & Wake, 2007; Adams & Nistri, 2010; or
head shape: Adams, 2004, 2010). Similarly, aspects of
shape morphology that are not captured in one of the six
PCs but are in fact the result of ecomorphological
adaptations might not be identified as such. This is a
major caveat in the use of individual PCs to describe
morphological variation, as compared to using
approaches such as factor analysis (for a discussion of
factor analysis in morphometrics see Blackith & Reyment, 1971; Mitteroecker & Bookstein, 2007). However,
since these PCs were identified by PCA as the major
gradients of variation, it seems unlikely that morphological adaptations to microhabitats have occurred in other
directions in morphospace that are not represented by
these PCs. Finally, even if there is important ecomorphological variation in other traits (or in other ways of
looking at these same traits), the generally weak relationship between microhabitat and the major body
proportions that we included is still remarkable.
The lack of relationship between morphological variation and microhabitat use in temperate plethodontids
is most surprising when contrasted with lizards (nonsnake squamate reptiles). Most lizards have body forms
that are similar to those in salamanders and many
studies show strong relationships between morphology
and habitat (e.g. Losos, 1990a,b; Vitt et al., 1997;
Irschick & Losos, 1998; Glor et al., 2003; Wiens et al.,
2006; but see Vanhooydonck & Van Damme, 1999;
Schulte et al., 2004). Why are similar relationships weak
or absent in plethodontids? Below, we suggest a few
tentative hypotheses, but we acknowledge that these
are highly speculative and will require additional work
to evaluate.
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
644
T. BLANKERS ET AL.
One hypothesis is that the strong relationships between morphology and microhabitat use in lizards are
not replicated in salamanders because of major differences in behaviour and ⁄ or physiology between lizards
and salamanders. For example, both laboratory and field
studies in lizards show that longer limbs increase maximal sprinting speed and acceleration, and facilitate
predator escape behaviour (Irschick & Jayne, 1998;
Irschick & Losos, 1998; Irschick, 2002). However, it is
not clear that plethodontids generally sprint away from
predators as lizards do. For example, many Plethodon
exhibit an immobile posture when disturbed, presumably
to be less conspicuous to visually oriented predators (e.g.
Dodd, 1990). Further, plethodontids are typically slow
moving and have low metabolic rates relative to other
amphibians and reptiles (Vitt & Caldwel, 2009). Thus,
plethodontids may generally be more homogeneous in
their locomotor performance in different environments.
Alternately, plethodontids may differ significantly in
their performance in different environments due to
differences in behaviour and physiology, but without
corresponding differences in morphology. These two
hypotheses could be tested by comparing the performance of plethodontid species that occur in different
environments at different ecologically relevant activities
(e.g. climbing, terrestrial movement, swimming). If the
first hypothesis is correct, species with similar morphology should have similar performance at these different
activities, regardless of the microhabitat in which they
typically occur. If the second hypothesis holds, there
should be strong relationships between the microhabitat
of species and their performance at these activities (e.g.
arboreal species should be superior climbers, aquatic
species should be superior swimmers), even if there are
few morphological differences among species.
A related hypothesis that might explain the weak
relationship between morphology and microhabitat is
that there are developmental constraints on morphological evolution in plethodontids which limit their adaptation to different microhabitats. In fact, plethodontids are
well known for their general morphological conservatism
(e.g. rates of morphological evolution are decoupled from
rates of species diversification; Adams et al., 2009). We
can think of two potential examples of how developmental constraints might limit ecomorphological evolution in plethodontids, although both examples are also
somewhat problematic. First, vertebral number is fixed in
most tropical bolitoglossines (Wake, 1966), which may
constrain their evolution of body shape (but see ParraOlea & Wake, 2001). However, we note that tropical
bolitoglossines also show the strongest relationships
between morphology and microhabitat in our study.
Second, species with an aquatic larval stage and a
terrestrial adult stage must be able to move both in
water and on land. Given that the basic body form is
established prior to metamorphosis in plethodontids (e.g.
head, trunk, limb, and tail proportions), then the need
for the same body shape to function on both land and in
water may constrain adaptation of adult morphology to
particular microhabitats. However, it should be noted
that most plethodontines (excluding most Desmognathus)
and all bolitoglossines have direct development and
therefore lack an aquatic larval stage (Vitt & Caldwel,
2009). In summary, plethodontids show surprisingly
little relationship between morphology and microhabitat
categories, but determining the causes of this pattern will
require more detailed studies of their behaviour, performance, morphology, and development.
Acknowledgments
We thank the Stichting Dr. Hendrik Muller Vaderlandsch
Fonds, the Amsterdams Universiteits Fonds, and the
University of Amsterdam for providing scholarships that
facilitated T. Blankers’ trip to Stony Brook to conduct this
part of his graduate research project. For helpful comments that greatly improved the manuscript, we thank
R. Snook, G. Thomas, and an anonymous reviewer. For
permission to use photos we thank (in alphabetical
order): Matthew J. Aresco (Phaeognathus hubrichti),
Eduardo Boza (Nototriton richardi), Dante B. Fenolio
(Eurycea rathbuni), Michael Graziano (Plethodon wehrlei;
Stereochilus marginatus), Joyce Gross (Plethodon cinereus),
James Hanken (Pseudorycea naucampatepetl), Gary Nafis
(Plethodon vandykei), Vide Ohlin (Oedipina pacificensis),
Alan Resetar (Plethodon glutinosus), Sean M. Rovito
(Chiropterotriton multidentatus), Josiah H. Townsend (Bolitoglossa celaque), Jon Wedow (Bolitoglossa schizodactyla),
and John White (Aneides aeneus).
References
Adams, D.C. 2004. Character displacement via aggressive
interference in Appalachian salamanders. Ecology 85: 2664–
2670.
Adams, D.C. 2010. Parallel evolution of character displacement
driven by competitive selection in terrestrial salamanders.
BMC Evol. Biol. 10: 1–10.
Adams, D.C. & Nistri, A. 2010. Ontogenetic convergence and
evolution of foot morphology in European cave salamanders
(Family: Plethodontidae). BMC Evol. Biol. 10: 216–226.
Adams, D.C. & Rohlf, F.J. 2000. Ecological character displacement in Plethodon: biomechanical differences found from a
geometric morphometric study. Proc. Natl Acad. Sci. USA 97:
4106–4111.
Adams, D.C., Berns, C.M., Kozak, K.H. & Wiens, J.J. 2009. Are
rates of species diversification correlated with rates of morphological evolution? Proc. R. Soc. Lond. B 276: 2729–2738.
Akaike, H. 1983. Information measures and model selection. Int.
Stat. Inst. 44: 277–291.
AmphibiaWeb. 2011. AmphibiaWeb: Information on Amphibian
Biology and Conservation. http: ⁄ ⁄ amphibiaweb.org ⁄
Blackith, R.E. & Reyment, R.A. 1971. Multivariate Morphometrics.
Academic Press, London ⁄ New York.
Burnaby, T.P. 1966. Growth-invariant discriminant functions
and generalized distances. Biometrics 22: 96–110.
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
Salamander ecomorphology
Burnham, K.P. & Anderson, D.R. 2002. Model Selection and
Multimodel Inference: A Practical Information-Theoretic Approach.
Springer Science, New York.
Clabaut, C., Bunje, P.M.E., Salzburger, W. & Meyer, A. 2007.
Geometric morphometric analyses provide evidence for the
adaptive character of the Tanganyikan cichlid fish radiations.
Evolution 61: 560–578.
Dodd, C.K. Jr 1990. Postures associated with immobile woodland salamanders, genus Plethodon. Fla. Sci. 53: 43–49.
Gavrilets, S. & Losos, J.B. 2009. Adaptive radiation: contrasting
theory with data. Science 323: 732–737.
Glor, R.E., Kolbe, J.J., Powell, R., Larson, A. & Losos, J.B. 2003.
Phylogenetic analysis of ecological and morphological diversification in Hispaniolan trunk-ground anoles (Anolis cybotes
group). Evolution 57: 2383–2397.
Grafen, A. 1989. The phylogenetic regression. Phil. Trans. R. Soc.
Lond. B 326: 119–157.
Grant, P.R. 1999. Ecology and Evolution of Darwin’s Finches.
Princeton University Press, Princeton.
Harmon, L.J., Weir, J., Brock, C., Glor, R.E. & Challenger, W.
2008. GEIGER: investigating evolutionary radiations. Bioinformatics 24: 961–964.
Irschick, D.J. 2002. Evolutionary approaches for studying functional morphology: examples from studies of performance
capacity. Integr. Comp. Biol. 42: 278–290.
Irschick, D.J. & Jayne, B.C. 1998. Effects of incline on speed,
acceleration, body posture and hindlimb kinematics in two
species of lizard Callisaurus draconoides and Uma scoparia. J. Exp.
Biol. 201: 273–287.
Irschick, D.J. & Losos, J.B. 1998. A comparative analysis of the
ecological significance of maximal locomotor performance in
Caribbean Anolis lizards. Evolution 52: 219–226.
Irschick, D.J., Vitt, L.J., Zani, P.A. & Losos, J.B. 1997. A
comparison of evolutionary radiations in mainland and
Caribbean Anolis lizards. Evolution 78: 2191–2203.
IUCN. 2010. IUCN Red List of Threatened Species Version 2010.4.
[Database]. URL: www.iucnredlist.org.
Jaekel, M. & Wake, D.B. 2007. Developmental processes
underlying the evolution of a derived foot morphology in
salamanders. Proc. Natl Acad. Sci. USA 104: 20437–20442.
Jockusch, E.L. 1997. Geographic variation and phenotypic
plasticity of number of trunk vertebrae in slender salamanders, Batrachoseps (Caudata: Plethodontidae). Evolution 51:
1966–1982.
Jolicoeur, P. 1963. The multivariate generalization of the
allometry equation. Biometrics 19: 497–499.
Kozak, K.H., Larson, A., Bonett, R.M. & Harmon, L.J. 2005.
Phylogenetic analysis of ecomorphological divergence,
community structure, and diversification rates in dusky
salamanders (Plethodontidae: Desmognathus). Evolution 59:
2000–2016.
Kozak, K.H., Mendyk, R.W. & Wiens, J.J. 2009. Can parallel
diversification occur in sympatry? Repeated patterns of bodysize evolution in co-existing clades of North American salamanders. Evolution 63: 1769–1784.
Lombard, E. & Wake, D.B. 1986. Tongue evolution in the
lungless salamanders, family Plethodontidae. IV. Phylogeny of
plethondontid salamanders and the evolution of feeding
dynamics. Syst. Zool. 35: 532–551.
Losos, J.B. 1990a. Ecomorphology, performance capability, and
scaling of West Indian Anolis lizards: an evolutionary analysis.
Ecol. Monogr. 60: 369–388.
645
Losos, J.B. 1990b. The evolution of form and function: morphology and locomotor performance ability in West Indian
Anolis lizards. Evolution 44: 1189–1203.
Maddison, W.P. & Maddison, D.R. 2010. Mesquite: A Modular
System for Evolutionary Analysis. Version 2.74. http://mesquite
project.org.
Martins, E.P. & Hansen, T.F. 1997. Phylogenies and the
comparative method: a general approach to incorporating
phylogenetic information into the analysis of interspecific
data. Am. Nat. 119: 646–667.
McCoy, M.W., Bolker, B.M., Osenberg, C.W., Miner, B.G. &
Vonesh, J.R. 2006. Size correction: comparing morphological
traits among populations and environments. Oecologia 148:
547–554.
McCranie, J.R. & Wilson, L.D. 2002. The Amphibians of Honduras.
Society for the Study of Amphibians and Reptiles, Salt Lake
City, UT.
Mitteroecker, P. & Bookstein, F. 2007. The conceptual and
statistical relationship between modularity and morphological
integration. Syst. Biol. 56: 818–836.
Mosimann, J.E. 1970. Size allometry: size and shape variables
with characterizations of the lognormal and generalized
gamma distributions. J. Am. Stat. Assoc. 65: 930–945.
Mosimann, J.E. & James, F.C. 1979. New statistical methods for
allometry with application to Florida red-winged blackbirds.
Evolution 33: 444–459.
Orme, D., Freckleton, R., Thomas, G., Petzoldt, T. & Fritz, S.
2009. CAIC: comparative analysis using independent contrasts. URL: http://R-Forge.R-project.org/projects/caic/.
Orme, D., Freckleton, R., Thomas, G., Petzoldt, T., Fritz, S. &
Isaac, N. 2011. caper: comparative analyses of phylogenetics
and evolution in R. URL: http://cran.r-project.org/web/
packages/caper/index.html.
Pagel, M. 1999. The maximum likelihood approach to reconstructing ancestral character states of discrete characters on
phylogenies. Syst. Biol. 48: 612–622.
Paradis, E., Claud, J. & Strimmer, K. 2004. APE: analyses of
phylogenetics and evolution in R language. Bioinformatics 20:
289–290.
Petranka, J.W. 1998. Salamanders of the United States and Canada.
Smithsonian Institution Press, Washington, DC.
Pinheiro, J., Bates, D., DebRoy, S. & Sarkar, D. 2011. nlme:
Linear and nonlinear mixed effects models. URL: http://
cran.r-project.org/web/packages/nlme/index.html.
Rice, W.R. 1989. Analyzing tables of statistical tests. Evolution 43:
223–225.
Schluter, D. 2000. The Ecology of Adaptive Radiation. Oxford
University Press, Oxford, UK.
Schluter, D., Price, T., Mooers, A.A. & Ludwig, D. 1997.
Likelihood of ancestor states in adaptive radiation. Evolution
51: 1699–1711.
Schulte, J.A., Losos, J.B., Cruz, F.B. & Nuñez, H. 2004. The
relationship between morphology, escape behaviour and
microhabitat occupation in the lizard clade Liolaemus (Iguanidae: Tropidurinae*: Liolaemini). J. Evol. Biol. 17: 408–420.
Stacklies, W., Redestig, H., Scholz, M., Walther, D. & Selbig, J.
2007. pcaMethods – a biodoconductor package providing
PCA methods for incomplete data. Bioinformatics 23: 1164–
1167.
Vanhooydonck, B. & Van Damme, R. 1999. Evolutionary
relationships between body shape and habitat use in lacertid
lizards. Evol. Ecol. Res. 1: 785–805.
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY
646
T. BLANKERS ET AL.
Vitt, L.J. & Caldwel, J.P. 2009. Herpetology: An Introductory Biology
of Amphibians and Reptiles. Academic Press, San Diego.
Vitt, L.J., Caldwell, J.P., Zani, P.A. & Titus, T.A. 1997. The role of
habitat shift in evolution of lizard morphology: evidence from
tropical Tropidurus. Proc. Natl Acad. Sci. USA 94: 3828–3832.
Wainwright, P.C. & Reilly, S.M. 1994. Ecological Morphology:
Integrative Organismal Biology. University of Chicago Press,
Chicago.
Wake, D.B. 1966. Comparative osteology and evolution of the
lungless salamanders, family Plethodontidae. Mem. South. Calif.
Acad. Sci. 4: 1–111.
Wake, D.B. 1987. Adaptive radiation of salamanders in Middle
American cloud forests. Ann. Missouri Bot. Gard. 74: 242–264.
Wake, D.B. 2009. What salamanders have taught us about
evolution. Annu. Rev. Ecol. Evol. Syst. 40: 333–352.
Wiens, J.J. & Hoverman, J.T. 2008. Digit reduction, body size,
and paedomorphosis in salamanders. Evol. Dev. 10: 449–463.
Wiens, J.J., Brandley, M.C. & Reeder, T.W. 2006. Why does a
trait evolve multiple times within a clade? Repeated evolution
of snakelike body form in squamate reptiles. Evolution 60: 123–
141.
Wiens, J.J., Parra-Olea, G., Garcia-Paris, M. & Wake, D.B. 2007.
Phylogenetic history underlies elevational biodiversity patterns in tropical salamanders. Proc. R. Soc. Lond. B 274: 919–
928.
Zheng, Y., Peng, R., Kuro-o, M. & Zeng, X. 2011. Exploring
patterns and extent of bias in estimating divergence time from
mitochondrial DNA sequence data in a particular lineage: a
case study of salamanders (Order: Caudata). Mol. Biol. Evol. 28:
2521–2535.
Supporting information
Additional Supporting Information may be found in the
online version of this article:
Appendix S1 R codes.
Appendix S2 Supplementary material:
Table S1 Microhabitat assignment, geographic region,
and species means for the seven morphological variables
for 189 plethodontid salamander species.
Table S2 Comparison of AIC scores for the different
models of character evolution.
Table S3 Correlation coefficients of microhabitat categories for MicrohabitatPC5.
Table S4 Multivariate PGLS on relationship between
microhabitat use and morphology, using PC1–PC4 and
PC6 as response variable (i.e. excluding PC5).
Table S5 Relationship between microhabitat and sizeuncorrected morphological variables.
Table S6 Relationship between microhabitat and
morphology using the phylogeny with a root age of
48 Myr.
Table S7 Relationship between microhabitat and
morphology using the phylogeny with a root age of
69 Myr.
Table S8 Eigenvalues, percentage variance explained,
and loadings for the morphological variables for PC1–PC6
for the temperate plethodontid species.
Table S9 Relationship between microhabitat use
(scenarios 1–6) and morphology for temperate species.
Table S10 Relationship between microhabitat use
(scenarios 10–15) and morphology for temperate species.
Table S11 Eigenvalues, percentage variance explained,
and loadings for the morphological variables for the
tropical subgroup on PC1–PC6.
Table S12 Relationship between microhabitat use and
morphology for tropical species.
As a service to our authors and readers, this journal
provides supporting information supplied by the authors.
Such materials are peer-reviewed and may be reorganized for online delivery, but are not copy-edited or
typeset. Technical support issues arising from supporting
information (other than missing files) should be
addressed to the authors.
Received 27 September 2011; revised 7 December 2011; accepted 16
December 2011
ª 2012 THE AUTHORS. J. EVOL. BIOL. 25 (2012) 634–646
JOURNAL OF EVOLUTIONARY BIOLOGY ª 2012 EUROPEAN SOCIETY FOR EVOLUTIONARY BIOLOGY